top of page
TOSHI Assist+
Client
TOSHI (luxury last-mile logistics)
Team
Sydney Marcus (Design Lead)
Edwin Wills (CTO)
Margaret Klein (Operations Lead)
team included: developers, a core testing group & ops stakeholders
the problem.
Prior to my joining TOSHI, the delivery network (assistants) relied on a combination of three systems to be able to complete deliveries: an external app, Elogii, a custom web-app, and paper signature cards. Drivers had to switch frequently between the three throughout their day, leading to missing data, inefficiency in the network, long onboarding times (due to confusion in the process) and a poor end-customer experience at delivery (drivers returning to a customer to collect a forgotten signature, the wrong packages being delivered, uncomfortable interactions when the app stopped working due to low-signal).
TOSHI's core mission was to make end-customers and brand partners feel as much joy during the last-mile as they did during the purchase experience, and the assistant process posed a significant hurdle.
My Role
Research – ride alongs, interviews, shadowing/doing deliveries
Design – design system creation, app & system architecture, UX/UI
Testing – early concept reviews, user testing, QA
Duration
1st release - 5 months, March 2023
2nd release - 2 months, May 2023
Continuous iteration - Nov 2024
the solution.
An intuitive and user-friendly native driver app that integrates seamlessly into TOSHI's new cohesive eco-system (including booking platform, operations dashboard and customer-facing order tracking system) that makes each user of the eco-system feel like they are in the "driver's seat" of the experience.
The driver app eliminated the need for three separate apps as it included everything necessary to complete TOSHI's complex tasks, and was architected in a flexible and modular way that allowed for constant iteration, improvement and feature expansion. This app was also the first to benefit from the new design system, which was then applied to all platforms within the eco-system.
the impact.
Implementing the new TOSHI Assist app delivered measurable results:
-
Improved Driver Efficiency (JPH) – streamlined processes enabled faster deliveries, cutting average time at delivery from 7.25 minutes to 2.75 minutes
-
Customer NPS Increase – satisfaction scores from the end customer (not drivers) improved to 9.2 from 6.3, reflecting better service experiences
-
Operational Growth – business was able to easily scale and onboard an additional 50-drivers for peak in under 1 week across three cities (London, NYC and LA)
Post-launch, we continuously refined the app based on feedback as well as building new workflows:
-
Real-Time Notifications – enabled dynamic updates for better task management
-
Workflow Optimization – iterations reduced task complexity, further improving driver efficiency
-
Cost Savings – loss expenses decreased by 20% through better warehouse dispatching and scanning
Take a quick look at other parts of the TOSHI eco-system by clicking on the images below –
I have more files and details to share about these projects in interviews
business challenge.
How might we meet our OKRs by improving the driver experience?
Issues with the driver app and other digital systems at TOSHI not only negatively impacted TOSHI's drivers, operations team members and end-customers, but the main OKR that would lead to profitability as well – journeys per driver per hour (JPH).
In order to fully understand the current process and it's issues, I took a three pronged research approach – shadow (ride along with the drivers), interview (all stakeholders impacted by the driver app), and survey (set a benchmark of "ease of use" of the current system for comparison against any new ideas or solutions).
My research uncovered 3 main issues: inefficient workflows, lack of clarity and customer dissatisfaction. Collaborating with product, engineering and operations we brainstormed solutions to each of these issues that contributed to improving stakeholder experience and JPH, leading to the new app – TOSHI Assist.
process.
The process for this project was cyclical and involved a lot of iteration, but a brief overview of the steps are included here and broken down in more detail below:
01.
Research
-
Ride-alongs
-
Interviews
-
Benchmark survey
02.
Journey Map
-
Current journeys
-
Pain points
-
Potential future paths
03.
Site Architecture
-
Simple workflow
-
Clear navigation
-
Flexibility for future
04.
Prototype
-
Initial wireframes
-
Clickable low-fi
-
Multiple options
05.
Test
-
Continuous feedback
-
Live prototype testing
-
Sample driver routes
06.
Iterate
-
Revise and rebuild
-
Implement new paths
-
Co-design/build wkshp
07.
Design System
-
Design components
-
Build in storybook
-
Implement in app
08.
Evaluate
-
Measure against benchmark
-
Regular testing/usage
research.
As mentioned above I utilized a three pronged approach to the research to ensure I had the quality and quantity of feedback to capture a representative sample of the diverse experiences of our stakeholders with as much depth as possible. The research was focused around the following questions:
-
Where is the current system failing?
-
Where is the current system succeeding?
-
How do we build for the future instead of the current?
-
How is the driver process impacting the other stakeholders in the process?
-
How is the driver process impacting our OKR of three journeys per hour (JPH)?
Ride Alongs
5 days in London and 3 in NYC
Interviews
12 drivers, 4 ops team, 15 customers
Benchmarking Survey
All 20 drivers and 6 ops completed
Key pain points:
-
Inefficient workflows – manual processes are slowing deliveries (average time at delivery was 7.25 minutes)
-
Lack of clarity – drivers struggled with incomplete order information and confusing in-app navigation
-
Customer dissatisfaction – delays and miscommunications are impacting customer reviews of the service (NPS at 6.3)
Positive elements:
-
When an address was selected within the 3rd party app (Elogii), it directed auutomatically to Waze, and the car's GPS picked up the route
-
Driver was able to call the customer directly from the 3rd party app (Elogii) if no one responded to the knock at the door
-
Signature "on glass" was a very popular feature with drivers and customers (when it functioned properly)
Outcomes:
-
Identification of core features – order information, navigation, adjusting items at collection, signature "on-glass" via the app, feedback
-
Potential points for time saving – pre-populating customer information at signature, quick links for navigation, etc
-
Impacts on other stakeholders – customers negatively impacted by issues with app (potential revenue impact), significant operational team member time lost manually transitioning journeys (major cost impact), tech team time lost trying to recover missing data (major cost impact)
journey mapping & site architecture.
Focused on architecting an app that addressed the pain points of current flows, but was built with the flexibility to grow with the business, our team re-imagined the steps of the existing journey and divided it into three stages – assignment, collection and drop off. This new model simplified the data and allowed for all journeys to be treated the same independent of their "origin location", opening TOSHI up to new flows such as returns, B2B journeys and omnichannel fulfillment.
-
Simple workflows
-
Clear navigation
-
Flexibility for future
Bottlenecks:
-
Order Acceptance: drivers lacked clarity on task specifics
-
Routing: requires seamless integration & real-time updates
-
Customer Communication: manual updates caused delays and errors
The journey map pinpointed areas for improvement, such as dynamic task updates, streamlined navigation, and automated status alerts
A simplified structure:
-
Dynamic dashboard where drivers access real time updates
-
Streamlined navigation where information and notifications are presented in a prioritized and digestible format
-
Integrated features including routing tools and a calling feature centralized into a single interface
This architecture allowed for task consolidation and reduced cognitive load for drivers
prototype, test, iterate.
Low-Fidelity Prototypes
Initial wireframes were created to outline the app’s core workflows, focusing on clarity in task management and navigation. These prototypes emphasized function over form, helping drivers visualize task sequences and information display. Key stakeholders and a small group of drivers reviewed these wireframes, providing feedback on the basic structure and usability.
High-Fidelity Prototypes – Implementing the new Design System
With wireframe feedback integrated, high-fidelity prototypes were built to closely mimic the final product. These prototypes featured polished UI elements from the newly developed design system, and dynamic components such as a task prioritization dashboard. Interactive prototypes allowed drivers to simulate completing deliveries, from accepting orders to marking tasks as done.
User Testing
-
Moderated Usability Tests: drivers were observed completing simulated deliveries, highlighting pain points like confusion over notifications and unclear button placements
-
Feedback Sessions: detailed feedback was collected from drivers and stakeholders about user flows, visual hierarchy, and ease of navigation
-
Quantitative Analysis: metrics like task completion time, error frequency, and user satisfaction were tracked, guiding design improvements
Iterative Refinement – each round of testing informed design updates
-
Task Flows: steps were streamlined to minimize clicks and improve task clarity
-
Visuals: button sizes, color contrasts, and icon placements were optimized based on user input
-
Real-Time Updates: enhancements like priority task flags and route recalculation improved workflows
Post-Launch Iterations – user feedback after release led to further refinement
-
Dynamic Notifications: real-time adjustments for route changes were implemented
-
Driver Feedback Loop: an in-app survey allowed drivers to submit improvement suggestions directly
Iterations of the dashboard landing page, from existing web app, to V3 of the native app
data driven decisions.
Leveraging data to guide design priorities and decision making was essential to achieving buy-in across the business as well as ensuring our designs were user-friendly and aligned with measurable outcomes. Using user research and cyclical feedback allowed us to pinpoint inefficiencies, validate design hypotheses, and prioritize impactful changes. By grounding decisions in data, we achieved tangible results such as faster task completion, improved customer NPS, and higher JPH, demonstrating the value of actionable insights in driving meaningful design outcomes.
Data
Decision
Description
-
Drivers going "out-of-order" led to a 15% increase in drive time between locations
-
Approx. 3 hours of ops team time per day spent troubleshooting routes
Dashboard clearly indicates "next task" and disables future tasks
App dashboard forces tasks to be completed in the optimal order as decided by the TOSHI plan routing system. Ops team can visualize the routes using a clear, map-based interface for quicker resolution.
-
5 drivers per week drove to a location that was already removed from their route – losing a minimum of 25 minutes per instance
-
17 drivers complained about not being notified of new tasks – high frustration and lack of trust
Real-time notifications and limits on re-routing
Dynamic, prioritized in-app notifications to alert drivers immediately about route changes and urgent tasks. Limit on routing algorithm so it cannot remove a task when the driver marks they are en-route.
-
25% error rate during task completion due to excessive steps and confusing instructions
Simplified task flows and reduced clicks per task
Redesigned workflows to reduce number of clicks for critical actions by 40% and lower error rate to 8%.
-
16 drivers indicated the original app was missing essential delivery information (via the survey)
Include essential info and improve data display
Improved data summary on dashboard, and enhanced information display throughout the app.
-
Route completion times revealed significant delays during multi-stop deliveries
-
Driver wait times at warehouses were very long and collections were inefficient
Optimize timing at multi-stops and warehouses
Introduced route clustering so multi-stop routes stack on the dashboard, reducing clicks and time. For the warehouse, an optimized queue system with live updates was added, cutting wait times by 30%.
-
Approx. 2 signature cards were "lost" per week
-
Data was often manually entered incorrectly, leading to high operational errors
Manual data errors
Eliminated signature cards, implemented "sign-on-glass" at 100% functionality. Added automated data validation, reducing ops errors by 25%.
visual design assets.
After rounds of prototyping, testing and iterating, I refined the app designs into the high-fidelity video and images shown below. The app remained in progress, with new features being added monthly up until TOSHI went into liquidation.
impact and results.
The TOSHI Assist app delivered substantial benefits across operational efficiency, customer satisfaction, and business scalability:
-
Enhanced Driver Efficiency: streamlined workflows and optimized route management enabled faster deliveries and reduced errors
-
Customer Satisfaction: the app’s user-centered design improved customer experience, reflected by a Net Promoter Score (NPS) increase to 9.2 from 6.3
-
Business Growth: scaled from one city to three (London, New York, and Los Angeles) and onboarded an additional 50 drivers
-
Cost Savings: optimized app portion of warehouse dispatch process reducing
loss expenses by 20% -
Design System Success: the cross-platform design system cut build times by 32% and ensured consistency across products, enabling faster feature delivery
Post-launch, we continuously refined the app based on feedback as well as building new workflows:
-
Real-Time Notifications – enabled dynamic updates for better task management
-
Workflow Optimization – iterations reduced task complexity, further improving driver efficiency
-
Cost Savings – loss expenses decreased by 20% through better warehouse dispatching & scanning
what I learned.
Data-driven decision making enhances impact
Delivery metrics, customer feedback, and NPS scores guided design priorities and validated decisions. This experience highlighted how aligning design updates with quantifiable insights can lead to measurable improvements in both user experience and operational performance.
Users should always be at the heart
Observing drivers in their environment highlighted the importance of designing for real-world usability. Empathy and deep understanding of user needs are crucial for creating effective solutions.
Collaboration strengthens design
Cross-functional collaboration between product, tech, and ops ensured alignment and balance between business goals and user needs. Early and regular involvement of all was key to our success.
Iterative testing builds robust products
Multiple testing cycles refined the app significantly, and prototyping and user feedback highlighted usability gaps early, minimizing costly post-launch fixes.
Scalability must be built-in
Creating a design system enabled rapid scalability across cities, products and user groups, demonstrating the importance of future-proofing during design and development.
Communication is critical
Pitching concepts to leadership and facilitating discussions with stakeholders contributed to my skills in articulating design decisions and rallying support for innovative solutions.
bottom of page